1. Field of the Invention
This invention relates to a liquefaction apparatus, and more particularly relates to an offshore apparatus for liquefying natural gas.
2. Description of the Related Art
Natural gas is obtained from gas, gas/condensate and oil fields occurring in nature, and generally comprises a mixture of compounds, the most predominant of which is methane. Usually, natural gas contains at least 95% methane and other low boiling hydrocarbon (although it may contain less); the remainder of the composition comprises mainly nitrogen and carbon dioxide. The precise composition varies widely, and may include a variety of other impurities including hydrogen sulphide and mercury.
Natural gas may be "lean" gas or "rich" gas. These terms do not have a precise meaning, but it is generally understood in the art that a lean gas will tend to have less higher hydrocarbons than a rich gas. Thus, a lean gas may contain little or no propane, butane or pentane, whereas a rich gas would contain at least one of these materials.
Since natural gas is a mixture of gases, it liquifies over a range of temperatures; when liquefied, natural gas is called "LNG" (liquefied natural gas). Typically, natural gas compositions will liquefy, at atmospheric pressure, in the temperature range -165.degree. C. to -155.degree. C. The critical temperature of natural gas is about -90.degree. C. to -80.degree. C., which means that in practice it cannot be liquefied purely by the application of pressure it must be also be cooled below the critical temperature.
Natural gas is often liquefied before being transported to its point of end use. Liquefaction enables the volume of natural gas to be reduced by a factor of about 600. The capital costs, and running costs, of the apparatus required to liquefy the natural gas is very high, but not as high as the cost of transporting unliquefied natural gas.
The liquefaction of natural gas can be carried out by cooling the gas in countercurrent heat exchange relationship with a gaseous refrigerant, rather than with the liquid refrigerants used in conventional liquefaction methods, such as the cascade or propane-precooled mixed refrigerant processes. At least part of the refrigerant is passed through a refrigeration cycle which involves at least one compression step and at least one expansion step. Before the compression step, the refrigerant is usually at ambient temperature (ie the temperature of the surrounding atmosphere). During the compression step, the refrigerant is compressed to a high pressure, and is warmed by the compression process. The compressed refrigerant is then cooled with the ambient air, or with water if there is a water supply available, to return the refrigerant back to ambient temperature. The refrigerant is then expanded in order to cool it further. There are basically two methods of achieving the expansion. One method involves a throttling process, which may take place through a J-T valve (Joule-Thomson valve), wherein the refrigerant is expanded substantially isenthalpically. The other method involves a substantially isentropic expansion, which may take place through a nozzle, or, more usually, through an expander or turbine. The substantially isentropic expansion of the refrigerant is known in the art as "work expansion". When the refrigerant is expanded through a turbine, work may be recovered from the turbine: this work can be used to contribute to the energy required to compress the refrigerant.
It is generally recognised that work expansion is more efficient than throttling (a greater temperature drop can be achieved for the same pressure reduction), but the equipment is more expensive. As a result most processes usually use only work expansion, or a mixture of work expansion and throttling.
When natural gas of a particular composition is cooled at a constant pressure, then for any given temperature of the gas there will be a particular value for the rate of change of enthalpy (Q) of the gas. The temperature (T) can be plotted against Q to produce a "cooling curve" for natural gas. The cooling curve is highly dependent upon pressure: if the pressure is below the critical pressure, then the T/Q cooling curve is highly irregular, ie, it contains several portions of different gradient, including a portion of zero, or close to zero, gradient. With increases in pressure, particularly above the critical pressure, the T/Q cooling curve tends towards a straight line.
Reference is now made to FIG. 1, which is a graph of temperature vs. rate of change of enthalpy for the cooling of natural gas below and above critical pressure. The curve A, which is for the cooling of natural gas below critical pressure, will be considered in more detail. The curve A has a characteristic shape, which can be divided into a number of regions. Region 1 has a constant gradient and represents the sensible cooling of the gas. Region 2 has a decreasing gradient and is below the dew point temperature of the gas as heavier components begin to condense. Region 3 corresponds to the bulk liquefaction of the gas and has the lowest gradient in the curve: the curve in this portion is almost horizontal. Region 4 has an increasing gradient and is above the bubble point temperature of the liquid as the lightest components are condensed. Region 5 is below the bubble point temperature and is of a constant gradient, which is greater than the gradient of regions 3 and 4. Region 5 corresponds to the sensible cooling of the liquid; this is known as the "sub-cooling" region.
Reference is now made to FIG. 2 of the drawings, which is a graph of T/Q showing the combined cooling curve for natural gas and nitrogen, for a natural gas pressure of about 5.5 MPa. The graph also shows the warming curve for nitrogen over the same temperature range. This graph is representative of a liquefaction system in which natural gas is cooled in a series of heat exchangers by a simple nitrogen expander cycle. The nitrogen refrigerant exiting the series of heat exchangers is compressed, cooled with ambient air, cooled to about -152.degree. C. by work expansion, then fed to the cold end of the series of heat exchangers. The nitrogen refrigerant is precooled, before work expansion, by being passed through at least one heat exchanger at the warm end of the series of heat exchangers; thus, the cooling curve is a combined natural gas/nitrogen cooling curve.
The gradient of the cooling and warming curves at any particular point in FIG. 2 is dT/dQ. It is well known in the liquefaction field that the most efficient process is one which, for any given value of Q, the corresponding temperature on the cooling curve of the natural gas is as close as possible to the corresponding temperature on the warming curve of the refrigerant. This has the implication that dT/dQ for the cooling curve of the natural gas is as close as possible to dT/dQ for the warming curve of the refrigerant. However, for any given Q, the closer the temperature of the natural gas and the refrigerant, the higher the surface area needed for the heat exchanger. Thus, there has to be a certain trade off between minimising the temperature difference, and minimising the heat exchanger surface area. For this reason, it is generally preferred that for any given Q, the temperature of the natural gas is at least 2.degree. C. higher than that of the refrigerant.
In FIG. 2, the nitrogen warming curve is approximately a single straight line (ie, it has constant gradient). This is representative of a single stage refrigeration cycle, wherein the all the refrigerant nitrogen is cooled by work expansion to a low temperature of about -160.degree. C. to -140.degree. C., and is then passed in countercurrent heat exchange relationship with the natural gas. It is clear that at most parts of the T/Q curve there is a large temperature difference between the natural gas and the nitrogen refrigerant, and this indicates that the heat exchange is highly inefficient.
It is also known that the gradient of the warming curve of the refrigerant can be altered by changing the flow rate of the refrigerant through the heat exchangers: specifically, the gradient can be increased by decreasing the refrigerant flow rate. In the system shown in FIG. 2 it is not possible to decrease the nitrogen flow rate, because the increase in gradient will cause the nitrogen warming curve to intersect with the natural gas cooling curve. An intersection of the two curves is indicative of a temperature "pinch" or "cross-over" in the heat exchanger between the nitrogen and the natural gas, and under this condition it is impossible for the process to work.
However, if the nitrogen flow is split into two streams it is possible to make the nitrogen warming curve change from a single straight line into two intersecting straight line portions of different gradient. An example of such a process is disclosed in U.S. Pat. No. 3,677,019. This specification discloses a process in which the compressed refrigerant is split into at least two portions, and each portion is cooled by work expansion. Each work expanded portion is fed to a separate heat exchanger for cooling the gas to be liquefied. This causes the refrigerant warming curve to comprise at least two straight line portions of different gradient. This aids in the matching of the warming and cooling curves and improves the efficiency of the process. This specification was published over twenty years ago, and the process disclosed therein is inefficient by modern standards.
In U.S. Pat. No. 4,638,639 there is disclosed a process for liquefying a permanent gas stream, which also involves splitting the refrigerant stream into at least two portions in order to match the cooling curve of the gas to be liquefied with the warming curve of the refrigerant. The outlet of all the expanders in this process is at a pressure above about 1 MPa. The specification suggests that such high pressures increase the specific heat of the refrigerant, thereby improving the efficiency of the refrigerant cycle. In order to realise an efficiency improvement it is necessary for the refrigerant to be at, or near, its saturation point at the outlet of one of the expanders, because the specific heat is higher near to saturation. If the refrigerant is at the saturation point, then under these conditions there will be some liquid in the refrigerant that is fed to the heat exchangers. This leads to additional expense, because either the heat exchanger needs to be modified in order to handle a two-phase refrigerant, or the refrigerant needs to be separated into liquid and gaseous phases before being fed to the heat exchanger.
U.S. Pat. No. 4,638,639 is primarily concerned with processes in which the refrigerant comprises a portion of the gas to be liquefied, ie the refrigerant is the same as the gas to be liquefied. The specification is particularly concerned with a system in which nitrogen is liquefied using a nitrogen refrigerant. The specification does not specifically disclose a process in which natural gas is cooled by nitrogen, nor would it be expected to be useful in such a process, because all modern large-scale processes for liquefying natural gas use a mixed refrigerant cooling cycle. Furthermore, in U.S. Pat. No. 4,638,639 the gas being liquefied is cooled to a temperature just below its critical temperature. A series of three J-T valves are provided to sub-cool the gas being liquefied.
The earliest refrigerant cycle used for the liquefaction of natural gas was the cascade process. Natural gas can be cooled in the cascade process by successive cooling with, for example, propane, ethylene and methane refrigerants. The mixed refrigerant cycle, which was developed later, involves the circulation of a multi component refrigerant stream, usually after precooling to -30.degree. C. with propane. The nature of the mixed refrigerant cycle is such that the heat exchangers in the process must routinely handle the flow of a two phase refrigerant. This requires the use of large, specialised heat exchangers. The mixed refrigerant cycle is the most thermodynamically efficient of the previously known natural gas liquefaction processes: it enables the warming curve of the refrigerant to be closely matched to the cooling curve of the natural gas over a wide temperature range. Examples of mixed refrigerant processes are disclosed in U.S. Pat. Nos. 3,763,658 and 4,586,942, and in European Paten No 87,086.
One of the reasons for the widespread use of the mixed refrigerant cycle in the cooling of natural gas is the efficiency of that process. The installation of a typical mixed refrigerant liquefaction plant for natural gas would cost upward of $US 1,000,000,000, but the high cost can be justified by the efficiency gains. In order to be cost effective through economy of scale the mixed refrigerant plants typically need to be able to produce at least 3 million tonnes of LNG per annum.
The size and complexity of mixed refrigerants liquefaction plants is such that, to date, they have all been constructed, and located, on land. Due to the size of natural gas liquefaction plants, and the requirement for deep water harbours, they cannot always be located near to the natural gas fields. Gas from the natural gas fields is usually transported to the liquefaction plant by pipeline. In the case of offshore natural gas fields, there are severe practical limitations on the maximum length of the pipeline. This means that offshore natural gas fields that are more than about 200 miles from land are seldom developed.